Abstract:

Mangroves have critical ecological functions and social and economic value, and are an important target for protection in the coastal wetland ecosystem. By monitoring long-term dynamic changes in mangrove ecosystems, the overall change process can be systematically and accurately recorded, providing data support and a basis for decision-making on scientific protection and effective management of the ecosystem. This study focuses on mangrove forests in coastal areas of Guangdong Province. A map of the mangrove forest, from 1986 to 2018, was made using Landsat remote sensing image based on Google Earth Engine (GEE),which is a cloud computing platform. The Random Forest (RF) method was used to extract mangrove trees from 32 periods from 1986 to 2018, in Guangdong Province. The interannual variation in mangrove characteristics in coastal cities of Guangdong province were compared. In addition, the evolution characteristics of mangrove patches in Guangdong province were analyzed. The results show that 1) The computing capacity and massive data of the GEE cloud platform provide data support for analyzing the inter-annual evolution of mangroves in Guangdong province, which greatly improves the computing efficiency. From 1986 to 2018, the overall classification accuracy of mangrove remote sensing was higher than 90%, with high classification accuracy and reliable results. In general, the coastal mangrove area of Guangdong province first decreased and then increased, and the range of change gradually declined after 2014, remaining at about 11 000 hm². Mangrove forests are unevenly distributed in the province and occur mainly in the west. 2) Concerning coastal cities, mangroves are distributed in 14 cities, among which Zhanjiang and Yangjiang have the largest mangrove area, which is about 70% of the mangrove area of Guangdong province. The mangrove area changes in each city fall under three categories: decreasing first and increasing later, increasing fluctuation, and no obvious change. 3) From 1986 to 2018, the overall number of patches in mangrove forests in Guangdong province showed a decreasing trend, but the average patch area (MPS) showedan increasing trend, and mangrove fragmentation was reduced. In 2018, the mangrove MPS was 4.11 hm² in Guangdong province, and the total number of patches was 2 782. From 1986 to 2018, when the change trend of MPS in mangrove forests in Guangdong province was opposite to that of patch quantity, the changes of patches were mainly expansion and fragmentation. When MPS change trend was consistent with the change trend of plaque number, the increase and decrease in the change of plaque were dominant. Information on annual mangrove area distribution and structural changes can provide more detailed data and reference for the rational development and protection of mangroves and support ecological restoration and finely tuned mangrove management.

Key words: mangrove, inter-annual change monitoring, Google Earth Engine (GEE), Random Forest, Guangdong Province